University of South Carolina Scholar Commons eses and Dissertations 2018 Mitigating Advocacy Bias: e Effect Of e Reviewer Role On Tax Professional Judgment Mary E. Marshall University of South Carolina Follow this and additional works at: hps://scholarcommons.sc.edu/etd Part of the Business Administration, Management, and Operations Commons is Open Access Dissertation is brought to you by Scholar Commons. It has been accepted for inclusion in eses and Dissertations by an authorized administrator of Scholar Commons. For more information, please contact [email protected]. Recommended Citation Marshall, M. E.(2018). Mitigating Advocacy Bias: e Effect Of e Reviewer Role On Tax Professional Judgment. (Doctoral dissertation). Retrieved from hps://scholarcommons.sc.edu/etd/4665
73
Embed
Mitigating Advocacy Bias: The Effect Of The Reviewer Role ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
University of South CarolinaScholar Commons
Theses and Dissertations
2018
Mitigating Advocacy Bias: The Effect Of TheReviewer Role On Tax Professional JudgmentMary E. MarshallUniversity of South Carolina
Follow this and additional works at: https://scholarcommons.sc.edu/etd
Part of the Business Administration, Management, and Operations Commons
This Open Access Dissertation is brought to you by Scholar Commons. It has been accepted for inclusion in Theses and Dissertations by an authorizedadministrator of Scholar Commons. For more information, please contact [email protected].
Recommended CitationMarshall, M. E.(2018). Mitigating Advocacy Bias: The Effect Of The Reviewer Role On Tax Professional Judgment. (Doctoral dissertation).Retrieved from https://scholarcommons.sc.edu/etd/4665
LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES ......................................................................................................... viii
CHAPTER 1: INTRODUCTION ........................................................................................1 CHAPTER 2: THEORY AND HYPOTHESIS DEVELOPMENT ....................................7
Figure 4.1 The Reviewer Role as a Moderator of Advocacy Bias ....................................47
1
CHAPTER 1
INTRODUCTION
Tax professionals are required to exhibit both objectivity in their analyses and
advocacy for their clients (Internal Revenue Service 2014; AICPA 2010). Importantly,
absent objectivity, tax professionals’ advocacy attitudes can cause them to
inappropriately assign greater weight to evidence supportive of their clients’ preferred
positions (Bobek, Hageman, and Hatfield 2010; Kahle and White 2004). This departure
from objectivity is referred to as advocacy bias, which occurs when “tax accountants’
advocacy attitudes produce confirmation bias when evaluating [evidence or] case law”
(Roberts 1998, 90). When present, advocacy bias affects multiple judgments of tax
professionals, including evidence search (Kadous, Magro, and Spilker 2008; Cloyd and
Spilker 1999), evidence evaluation (Kahle and White 2004; Johnson 1993), and bias
detection (Cuccia, Magro, and Whisenhunt 2017; Barrick, Cloyd, and Spilker 2004).
Advocacy bias can increase the likelihood of unexpected taxes, interest, and IRS
penalties for both the taxpayer and the tax professional signing the return (Hatfield 2000),
which is especially alarming because most clients expect tax professionals to focus on
accuracy rather than aggressiveness (Fleischman and Stephenson 2012; Collins, Milliron,
and Toy 1992).
Some argue advocacy bias is mitigated by the review process (Cuccia et al. 2017;
Hatfield 2001), which could happen if (1) reviewers exhibit individual characteristics
(e.g., experience, expertise) that decrease their susceptibility to bias or (2) elements of the
2
reviewer role (e.g., accountability or role expectations) facilitate more objective judgment
and decision making for those occupying the reviewer role. Tax reviewers may not be
aware of either their own or preparers’ advocacy bias because it is an unintentional (and
perhaps subconscious) bias that is stronger for those in higher ranks and with more
experience (Spilker, Stewart, Wilde, and Wood 2016; Bobek, Hageman, and Hatfield
2014; Bobek et al. 2010). Thus, reviewers may be just as susceptible to advocacy bias as
preparers despite increased experience. It is possible, however, that elements of the
reviewer role (e.g., accountability or role expectations) prompt professionals to focus
more on objectivity. Thus, the present study focuses on the effects of the reviewer role on
advocacy bias because individual characteristics are unlikely to mitigate bias and may
exacerbate bias.
Role theory (Biddle 1979) suggests elements of the reviewer role could mitigate
advocacy bias because individuals change how they make decisions based on the role
they occupy. According to Biddle (1979, 56), roles are a collection of “behaviors that are
characteristic of persons in a context.” My study examines how roles, and the
expectations associated with those roles, influence tax professional behavior. Barrick et
al. (2004) identify three expectations of tax professionals: accuracy, advocacy, and
feedback. Both preparers and reviewers are expected to achieve accuracy and advocacy,
but only those in the reviewer role are expected to provide feedback. I propose the
additional expectation of providing feedback may lead reviewers to approach evidence
evaluation differently than preparers. Specifically, this study predicts reviewers approach
engagements with a heightened focus on objectivity because they feel accountable for
meeting the expectations associated with reviewing (i.e., evaluation of preparer effort and
3
conclusion, assessment of evidence search and evaluation process, etc.). Prior literature
finds increased accountability leads to greater cognitive effort and reduced susceptibility
to cognitive biases (Tetlock 1999), thus I predict professionals occupying the reviewer
role are more likely to evaluate evidence objectively (e.g., with less bias) because of the
accountability associated with providing feedback to preparers.
Research has not yet examined whether the influence of advocacy bias changes
when tax professionals occupy the reviewer role (compared to when they occupy the
preparer role). Prior research suggests reviewers are unlikely to detect advocacy bias
without obvious cues (Cuccia, Magro, and Whisenhunt 2017; Barrick, Cloyd, and Spilker
2004; Hatfield 2000); however, past studies have focused on how reviewers
inappropriately evaluate biased preparer conclusions. Tax professionals, especially those
at the senior level, are often assigned to work as preparers on some engagements and as
reviewers on others, which suggests tax professionals can shift from one role to another.
The present study examines whether advocacy attitudes influence reviewer evaluation of
both the preparer’s conclusion and the associated evidence supporting the conclusions. In
addition, preparers in this study evaluate the same evidence and provide the same
judgments as reviewers, which allows me to compare the effects of advocacy attitudes on
the same evidence evaluation in both roles.
An online experiment was administered to 75 experienced tax professionals
recruited through a Qualtrics Panel. The study’s hypotheses are tested with a between-
participant experiment with three conditions. Participants act as either a reviewer or
preparer. In addition, the study manipulates whether participants assigned as reviewers
review a memo that (1) includes a preparer conclusion in favor of the client-preferred
4
treatment or (2) does not include a preparer conclusion. This additional manipulation
allows the study to fully isolate the effect of role by comparing reviewer judgments both
with and without the influence of a preparer conclusion.
Professionals determine the tax treatment (hobby or business) of a net loss
incurred in a bicycle racing activity, which the client prefers to treat as a business. In all
conditions, professionals are provided with partner instructions (indicating the partner has
no preferred outcome), client facts, a draft file memo documenting the preparer’s
research, and relevant tax authority. Professionals indicate whether the client facts
indicate business or hobby treatment based on the “hobby loss” rules of I.R.C. §183 and
Reg. Sec. 1.183-2. In addition, professionals in the reviewer (preparer) conditions are
asked the likelihood they would sign off on a conclusion (conclude) that the activity be
considered a business. Finally, they respond to items measuring client advocacy and
accountability.
Using ANOVA and regression analysis, I examine the effect of client advocacy,
role, and accountability on tax professionals’ evidence evaluation processes and
conclusions. Bias occurs when advocacy attitudes influence evidence search and/or
evaluation (e.g., if strong client advocacy attitudes lead a professional to inappropriately
assess evidence as supportive of the client’s preference). Results indicate that tax
professionals who occupy the reviewer role are less likely to exhibit advocacy bias than
those in a preparer role. Further, I find no evidence that reviewers’ evidence evaluation
processes are influenced by the preparer conclusion they view. However, I do find
evidence that accountability to the client influences preparers’ judgments. Specifically,
increasing accountability to the client decreases the likelihood of preparers exhibiting
5
advocacy bias, which provides initial evidence that firms may be able to mitigate
preparer advocacy bias by increasing felt accountability.
This study contributes to the literature in both theoretical and practical ways.
First, by examining the effect of the reviewer role on tax professional judgments, this
study adds to a growing literature (e.g., Cuccia et al. 2017; Barrick et al. 2004) studying
whether the review process mitigates advocacy bias. Whereas prior literature has
separately examined the effects of advocacy bias on both reviewer and preparer tasks,
this study isolates role by comparing how professionals in the role of reviewer evaluate
the same set of evidence as those in the role of preparer. Results provide initial evidence
that occupying the reviewer role mitigates advocacy bias.1 This study’s design examines
the effect of an existing firm mechanism (i.e., the review process) on the presence of
advocacy bias. The findings provide insight into the importance of the review process in
ensuring objective evidence evaluation as part of client recommendations.
Second, this study provides initial evidence of the effects of accountability on the
likelihood of tax preparers exhibiting advocacy bias. Specifically, preparers who feel
accountable to the client are less likely to exhibit advocacy bias, which provides insight
to firms regarding the importance of conveying the expectation that preparers are
accountable for their recommendations. Third, this study examines the effects of
accountability to multiple sources, including documenting different effects of
accountability to the client and firm. Interestingly, I find that accountability to the client
decreases the likelihood a preparer will exhibit advocacy bias but, surprisingly,
1 This study supports findings in prior audit studies which suggest occupying the reviewer role may improve objectivity in professional judgments (Reimers and Fennema 1999; Ricchute 1999; Libby and Trotman 1993).
6
accountability to the firm does not. Thus, I provide insights for future research about the
effect of accountability, especially when there are multiple accountability sources. For
example, auditors may also feel accountable to both the firm and the client (Gibbins and
Newton 1994). Future research could build on this study’s findings to disentangle the
effects of accountability to multiple sources on auditor and tax professional judgment.
Fourth, I find tax professionals are likely to assume the partner prefers to agree
with the client preference even when the partner explicitly specifies no preference. As a
result, professionals may feel pressure to find support for an outcome they
inappropriately assume the partner prefers, especially when in the preparer role. Thus,
firms can avoid unintentionally aggressive recommendations by explicitly and credibly
communicating information about preferences (or lack thereof). Finally, this study
empirically tests the effect of approval bias in professional judgment. Approval bias
occurs when professionals’ bias leads them to endorse other professionals’ potentially
biased judgments (Bazerman et al. 2002). Prior literature finds advocacy bias persists
across rank in the firm and experience levels; thus, the tax context provides one of the
most likely areas for approval bias to manifest. However, this study finds no evidence
that endorsing a preparer’s conclusion (e.g., signing off on a conclusion to support the
client’s preferred position) influences the likelihood of reviewer advocacy bias.
The remainder of this paper is organized as follows. Section 2 reviews the
literature on advocacy bias, role theory, and accountability, providing theoretical
development of the hypotheses. Sections 3 and 4 describe methods and results. Section 5
concludes.
7
CHAPTER 2
THEORY AND HYPOTHESIS DEVELOPMENT
2.1 Advocacy Bias
A tax professional has both the “right and the responsibility to be an advocate for
the taxpayer” (AICPA 2010); however, s/he also have a responsibility to “objectively
evaluate all relevant facts and tax authorities when preparing advice” (Kadous and Magro
2001, 453). Favoring a client’s preference supports a tax professional’s goal of being an
advocate but doing so without sufficient support of relevant authority may lead to
inappropriate recommendations. When advocacy attitudes “inhibit their abilities to
may unintentionally interpret facts and/or evidence in a manner that inappropriately
supports the clients’ preferences. Prior tax literature has termed this “advocacy bias.”2
Advocacy bias can lead to the recommendation of aggressive tax positions;3
however, aggressiveness is not a direct indicator of advocacy bias. Rather, advocacy bias
occurs when professionals’ advocacy attitudes influence the judgments they make during
evidence search and/or evidence evaluation (see Figure 2.1). Because advocacy bias is
2 Advocacy bias is similar to confirmation bias because it leads to biased evaluation; however, advocacy bias refers to bias towards the client’s preference rather than the professional’s initial opinion or judgment. Tax professionals exhibiting advocacy bias attempt to confirm client preferences rather than objectively determine the appropriate tax treatment. See Bobek et al. (2010) for a synthesis of client advocacy research. 3 The present study uses the term “aggressive tax positions” to describe positions that could be overturned if challenged in Tax Court. Recommending an aggressive position is not necessarily illegal, but when such a recommendation is the result of biased evidence evaluation, the tax professional and the client are likely taking on more risk than they intend.
8
generally an unconscious bias, it is more difficult to detect and control than
aggressiveness. Prior studies establish that tax professionals acting as preparers exhibit
advocacy bias throughout the research process, including searching for evidence (Kadous
et al. 2008; Cloyd and Spilker 1999) and evaluating evidence (Kahle and White 2004;
Johnson 1993). For example, prior research has shown tax professionals assign greater
weight to court cases that support the client’s preferred treatment (Johnson 1993) and
focus primarily on evidence most consistent with the client’s preferences (Kadous et al.
2008; Hatfield 2001; Cloyd and Spilker 1999).
Research also shows reviewers may exhibit advocacy bias when determining
whether to rely on preparer research memorandums (Hatfield 2001) and when detecting
bias in others’ judgments (Cuccia et al. 2017; Barrick et al. 2004). Cuccia et al. (2017)
find reviewers are likely to detect preparer bias only when the memo is linguistically
stylized to emphasize the evidence supporting the client-preferred outcome. Barrick et al.
(2004) examine whether reviewers are subject to advocacy bias when the client preferred
position does not have a “realistic possibility” of being upheld in court. The authors find
reviewers are more likely to rely on biased memos that incorrectly support an
unsupportable client preference than on biased memos that correctly support an
alternative. That is, when both memos are biased, reviewers are more likely to rely on
and incorporate the memo that achieves the client preference. However, the authors also
find some evidence that reviewers can detect preparer bias. Specifically, they find
supervisors (e.g., reviewers) request more rework and are less persuaded when staff
submit clearly biased memos supporting a client preference that cannot be objectively
supported than when they support unbiased memos supporting the same unsupportable
9
client preference. That is, when both memos recommend an unsupportable position,
reviewers are more likely to rely on the unbiased memo.4 Thus, supervisors (reviewers)
appear likely to detect advocacy bias when that bias leads to obviously inappropriate
recommendations (Barrick et al. 2004) or highly stylized presentation (Cuccia et al.
2017). However, advocacy bias is generally considered to be unintentional (Bobek et al.
2010), which suggest it is likely judgments affected by advocacy bias are not always
accompanied by stylized writing, obviously unsupported recommendations, or other
obvious cues.
The abovementioned prior studies have established reviewer advocacy bias by
assessing whether reviewers sign off on (i.e., agree with) preparer recommendations to
support aggressive client preferences. Although advocacy bias can contribute to evidence
evaluation that supports aggressiveness, aggressiveness is not a direct indicator of
advocacy bias. As a result, an increase in aggressiveness is not a precise test of whether
advocacy bias influences judgment. Research has not yet identified whether reviewer
judgments are driven by (1) reviewers’ own advocacy biases, (2) over-reliance on
preparers’ biased recommendations, or (3) reviewers’ appetites for aggressiveness. This
study follows Bobek et al. (2010) and examines bias more directly by focusing on
whether advocacy attitudes influence evidence evaluation rather than focusing on the
aggressiveness of a conclusion.
4 Barrick, et al. (2004) examine a scenario where client preference is clearly unsupportable because it does not have a “realistic possibility” of being upheld if challenged in court. Thus, tax professionals are unable to support the client’s preference (advocacy expectation) because doing so would not be supported by relevant authority (accuracy expectation). Following Bobek, et al. (2010), my study incorporates sufficient ambiguity that biased professionals could determine meeting both expectations is possible.
10
2.2 Mitigating Advocacy Bias with the Review Process
Previous research has identified multiple antecedents and consequences of
advocacy bias; however, only a few studies have examined whether there are processes in
place to mitigate advocacy bias effectively. Some argue the review process can mitigate
advocacy bias (Cuccia et al. 2017). This argument is reasonable because the review
process serves as a quality control mechanism. Circular 230, which governs behavior of
tax professionals practicing before the IRS, requires a tax professional to use “reasonable
care in engaging, supervising, training, and evaluating” a subordinate (IRS 2014, 19).
Barrick et al. (2004) indicate all tax professionals have two expectations in a research
engagement: (1) accuracy, which requires that they evaluate the evidence and determine
the appropriate conclusion, and (2) advocacy, which requires that they “marshal
evidence” to support their client’s preferences. Reviewers have a third expectation of
“providing corrective feedback necessary for novice professionals to learn from their
experiences” (Barrick et al. 2004, 2).5 Given the differences between the two roles, the
review process could mitigate advocacy bias if (1) professionals acting as reviewers
possess individual characteristics (e.g., experience or expertise) that facilitate
identification of alternative explanations, omissions, or inconsistencies; or (2) elements of
the reviewer role (e.g., increased accountability) prompt professionals to approach a task
differently than when they are not in that role.
One way the review process could mitigate advocacy bias is if tax reviewers
possess individual characteristics that facilitate more objective judgments and decrease
5 The reviewer role has a similar focus in an audit environment. Yip-Ow and Tan (2000) describe an audit reviewer’s identification of alternative explanations, omissions, or inconsistences in the preparer’s work as one of the primary purposes of the review process.
11
susceptibility to advocacy bias when compared to preparers. Professionals serving as first
reviewers generally possess increased expertise and exposure to more complex tax issues
than preparers (AICPA 2016); thus, one might expect reviewers to be more objective than
preparers. However, prior research shows client advocacy attitudes are stronger for
professionals with more experience or higher rank (Spilker, Stewart, Wilde, and Wood
2016; Bobek et al. 2010). Although reviewers may have more experience and/or expertise
than preparers, they may also have increased propensity to exhibit advocacy bias given
their stronger advocacy attitudes. This study does not expect reviewers’ individual
characteristics (e.g., experience and/or expertise) to mitigate advocacy bias. Motivated by
role theory (Biddle 1979) and research in accountability (Lerner and Tetlock 1999;
Gibbins and Newton 1994; Tetlock 1985), I examine another way in which the review
process may mitigate advocacy bias: the effect of occupying the reviewer role.
2.3 Role Theory and Accountability
I am unaware of research directly comparing similar judgments of tax
professionals occupying the reviewer and preparer roles; however, the reviewer role has
been studied in an audit context. Specifically, audit reviewers are more focused on
inconsistent information, produce more plausible alternative hypotheses, and are more
sensitive to source objectivity than audit preparers (Ricchute 1999; Reimers and Fennema
1999; Ismail and Trotman 1995; Libby and Trotman 1993). These studies present two
potential explanations for differences in behavior across roles. First, Reimers and
Fennema (1999) and Libby and Trotman (1993) argue that the preparer (i.e., the “initial
decision maker”) is focused more on producing an initial judgment, which prompts a
focus on systematic decision making and on defending his/her conclusion(s). In contrast,
12
the authors describe the reviewers as entering the process after the initial decision has
already been made, which prompts a focus on counterarguments and alternative
explanations because “information that is novel or unexpected receives relatively more
attention” (Libby and Trotman 1993, 563) from those who are reviewing an existing
decision. Thus, prior studies suggest the reviewer can be more objective because s/he is
less focused on developing a conclusion and more focused on reviewing the accuracy and
appropriateness of an existing conclusion. Second, prior literature also suggests the high-
quality decision-making behaviors of audit reviewers could be associated with an
expectation of accountability, which is defined as the need to justify decisions and/or
actions to another (Lerner and Tetlock 1999; Tetlock 1985). This study argues the
reviewer role produces different evidence evaluation processes than does the preparer
role, whether by shifting from systematic to more holistic decision making or by
increasing accountability.6
According to role theory, individuals occupy multiple roles and each role is
accompanied by role expectations, which are learned over time through socialization
(Ammeter, Douglas, Ferris, and Goka 2004; Frink and Klimoski 1998; Biddle 1979).
Roles are not necessarily jobs; roles represent a collection of “behaviors that are
characteristic of persons in a context” (Biddle 1979, 56). Tax professionals occupy
multiple roles (e.g., preparer, reviewer, signer, etc.), often simultaneously, with multiple
role senders (e.g., partner, client, IRS, etc.) who have differing expectations. For
6 Tax professionals may respond to occupying the reviewer role differently than auditors. Prior findings in audit are informative of professional behavior, but tax professionals work in a different environment. Auditors are expected to operate with professional skepticism, which could enhance objectivity. In contrast, tax professionals are required to be advocates, which could reduce objectivity if tax professionals exhibit advocacy bias. By examining tax professional behavior, this study examines the effect of role in an environment without an expectation of professional skepticism.
13
example, both preparers and reviewers are expected to strive for accuracy and advocacy,
whereas only reviewers are expected to provide feedback to subordinates (Barrick et al.
2004). Further, although the review process is assumed to provide learning opportunities
for a preparer (Gibbins and Trotman 2002), those at the reviewer level are expected to
have the expertise necessary to facilitate that learning. Because providing feedback
requires reviewers to focus on the quality of both the workpapers and the conclusions, I
expect reviewers to be more focused on reaching objective conclusions than professionals
who are not expected to provide feedback. Formally stated below, this study predicts
reviewers are less likely to exhibit advocacy bias because of the expectations associated
with being a reviewer.
H1: Tax professionals occupying the reviewer role are less likely to exhibit advocacy bias than those occupying the preparer role.
2.4 Accountability and Advocacy Bias
In addition to examining the effect of occupying the reviewer role on the
likelihood of exhibiting advocacy bias, this study also examines (1) the effect of role on
accountability and (2) the effect of accountability on tax professionals’ likelihood of
exhibiting advocacy bias. Role theory suggests “role occupants” feel accountable for the
perceived role expectations sent by “role senders” (authoritative persons) (Biddle 1979).
These role expectations are developed both formally and informally and are largely based
on norms in a workplace environment or process (Frink and Klimoski 1998). Tax
professionals are expected to act as client advocates and to complete their work with due
diligence and objectivity. These expectations come from the firm, the client, and the IRS
(e.g., the “role senders”), so all tax professionals have the same role senders regardless of
the role they occupy. However, role senders may assign different expectations to
14
professionals based on role. As discussed by Barrick, et al. (2004), tax preparers are
expected to achieve accuracy and advocacy and tax reviewers are expected to achieve
accuracy, advocacy, and feedback. Ammeter, et al. (2004) explain that if professionals
perceived differences in expectations, this could lead to differences in felt accountability.
It is unclear whether reviewers are expected to exhibit stronger advocacy or increased
accuracy compared to preparers; however, only reviewers are required to provide
feedback. Hence, this paper expects professionals to perceive an additional expectation to
provide feedback, which role theory predicts will lead to increased accountability to the
firm (Ammeter et al. 2004; Hall, Blass, Ferris, and Massengale 2004; Biddle 1979).
Formally stated below, I predict tax professionals have higher accountability to the firm
when occupying the reviewer role than when occupying the preparer role.
H2: Occupying the reviewer role increases a tax professional’s perceived accountability to the firm. The effect of the reviewer role on accountability to the client is less clear.
Whereas firms have different expectations for professionals occupying the reviewer and
preparer roles, it is unclear whether clients’ expectations differ across roles. Further, even
if clients do have different expectations, it is difficult to predict whether professionals
perceive these differences. Nevertheless, it is possible that professionals feel different
levels of accountability to their clients as they move into different roles. For example,
professionals may feel more responsible for client retention as they gain experience.
When combined with the stronger client advocacy associated with higher ranks, an
increase in responsibility for client retention suggests reviewers may feel increased
accountability to the client. In contrast, they may also feel more pressure to protect the
firm from litigation, which may shift their focus away from client preferences and toward
15
the firm’s expectations. It is also possible that reviewers have the experience necessary to
more effectively manage the tradeoff between achieving client preferences and
minimizing firm exposure. Thus, I pose the following research question to further explore
the relationship between role and accountability to the client.
RQ: Does occupying the reviewer role influence a tax professionals’ perceived accountability to the client? Prior research finds accountability influences many accounting decisions
including: effort duration in tax research (Cloyd 1997), extent and breadth of audit testing
(Asare, Trompeter, and Wright 2000), level of consensus among audit team members
(Johnson and Kaplan 1991), conservativeness of audit fraud risk assessments (Hoffman
and Patton 1997), and amount of justification provided by auditors (Koonce, Anderson,
and Marchant 1995). Prior research generally manipulates accountability by requiring
justification of decisions or by informing participants their work will be reviewed by
superiors. In contrast, the present study examines whether occupying the reviewer role is
a naturally occurring prime which increases accountability. If occupying the reviewer
role affects decision makers similarly to being expected to justify ones’ decisions, then
the review process could mitigate bias by increasing the accountability felt by those
performing the reviews.
However, an increase in accountability may not always improve the likelihood of
the review process to mitigate advocacy bias and can lead to suboptimal decision making
in some cases (Siegel-Jacobs and Yates 1994). Specifically, accountability influences
judgment differently depending on (1) whether the accountability source’s (i.e., the
entity/person to which the individual feels accountable) preferences are known; (2)
whether there are multiple sources of accountability; and (3) the type of accountability
16
(i.e. outcome or process). If the accountability source’s preference is known, individuals
may engage in attitude shifting or defensive bolstering (Hatfield 2000; Gibbins and
Newton 1994; Tetlock 1985).7 In a tax context, examples of attitude shifting include (1)
ensuring research findings agree with the partner’s expected outcome and (2) following
the partner’s preferred research approach. Thus, attitude shifting leads tax professionals
to work towards the accountability source’s expectation. In contrast, when the
accountability source’s preferences are not known, accountability can lead to increased
cognitive effort, which may increase objectivity. For example, if a tax partner does not
express an opinion, there is no explicit target to shift towards. Accountability is most
likely to decrease bias when the source’s preferences are not known because development
of an objective, balanced argument to support a position (rather than engaging in attitude
shifting or defensive bolstering) is more likely to occur when source preferences are
unknown.
In addition, the direction of attitude shifting and/or defensive bolstering are
difficult to predict in a tax setting because public accounting professionals face multiple
sources of accountability (e.g., partner and client) who may have conflicting preferences
(Gibbins and Newton 1994). For example, tax professionals may feel accountable both to
their clients because of an advocacy relationship and to their firms because of career and
reputational concerns. Because the client’s preference is usually for tax minimization,
strong feelings of accountability to the client could lead a professional to align his or her
evidence evaluation process with tax minimization (e.g., a more aggressive position).
7 Attitude shifting occurs when decision makers match their decisions to those of the audience to which they feel accountable (the “accountability source”). Defensive bolstering occurs when decision makers increase justification of their conclusion because it conflicts with the source’s preferences.
17
Prior research shows tax professionals’ evidence evaluation is often biased to agree with
client preferences (Bobek et al.2010; Kahle and White 2004; Johnson 1993), so strong
feelings of accountability to the client could further exacerbate advocacy bias. A similar
outcome could occur if the professional assumes the firm wants to meet the client’s
preferences. However, it is more likely that the professional would assume the firm
prefers objective and defensible evaluations because firms are focused on long-term
outcomes such as client retention and avoiding penalties. Such a focus would prompt a
more thorough and unbiased process rather than attitude shifting or defensive bolstering.
Because advocacy bias limits objectivity, focusing on building an objective and thorough
analysis could mitigate advocacy bias by increasing the cognitive effort dedicated to the
task (Gibbins and Newton 1994).
The type of accountability (i.e., process or outcome) also influences whether
accountability improves decision making (Siegel-Jacobs and Yates 1996).8 Prior research
shows outcome accountability may lead to dysfunctional behavior while process
accountability improves decision quality (Siegel-Jacobs and Yates 1996). Because tax
professionals may perceive outcome expectations from both the client and the firm,
increased accountability could hinder objectivity and potentially increase bias. Advocacy
bias can be described as using a biased evidence evaluation (e.g., process) to support a
client’s preferred tax position (e.g., outcome). I expect (1) accountability to the firm to
reduce a tax professional’s advocacy bias by focusing him/her on the quality of the
evidence evaluation process and (2) accountability to the client to increase a tax
8 Process accountability relates to the steps taken to arrive at a final conclusion. An example in a tax context is whether the professional follows the appropriate workflow. Outcome accountability relates to the final conclusion, for example the professional’s recommendation to the client.
18
professional’s advocacy bias by focusing him/her on the client’s preferred outcome.9
Formally stated below, this paper expects perceived accountability to the firm and the
client to influence the likelihood tax professionals will exhibit advocacy bias.
H3a: Accountability to the firm decreases the likelihood tax professionals will exhibit advocacy bias. H3b: Accountability to the client increases the likelihood tax professionals will exhibit advocacy bias.
In summary, I study whether the reviewer role mitigates advocacy bias by
examining whether (1) professionals occupying the reviewer role are less likely to exhibit
advocacy bias, (2) the reviewer role increases accountability, and (3) accountability
mitigates the effect of advocacy attitudes on evidence evaluation (see Figure 2.2). I
expect the reviewer role to mitigate advocacy bias because of the expectations associated
with the reviewer role. In addition, I examine whether occupying the reviewer role
increases accountability to both the firm and the client, which could also influence the
likelihood of exhibiting advocacy bias.
9 Outcome accountability to the firm is held constant by informing participants the partner has no a priori preference related to outcome. This is a reasonable assumption because, although firm leadership is interested in client retention, they are likely more interested in providing quality service to their client. In addition, outcome accountability to the client is held constant by informing participants of the client’s preference.
19
Figure 2.1. This figure illustrates the necessary relationship between advocacy attitudes and evidence evaluation to indicate advocacy bias. Aggressiveness, or the likelihood of reaching a potentially-unsupported conclusion, is not indicative of bias. Rather, advocacy bias is indicated when advocacy attitudes influence the evaluation of evidence. Thus, without advocacy bias, evidence evaluation will not differ based on advocacy attitudes; however, with advocacy bias, evidence evaluation could be influenced by advocacy attitudes.
Figure 2.2. This figure comprehensively depicts the predictions of the hypotheses.
Exte
nt to
whi
ch e
vide
nce
eval
uatio
n su
ppor
ts
clie
nt-p
refe
rred
pos
ition
Client Advocacy
The Effects of Advocacy Attitudes on Evidence Evaluation
No Advocacy Bias Advocacy Bias
20
CHAPTER 3
EXPERIMENTAL METHODS
3.1 Participants
Seventy-five experienced tax professionals currently working in public
accounting were recruited through a Qualtrics panel. The effect of reviewer role is
isolated by randomly assigning participants of similar rank and experience to each role,
so only participants who have current or recent (e.g., within two years) experience in both
the preparer and reviewer roles were allowed to continue through the screening process.
In addition, participants must work (1) in a public accounting firm, (2) in tax, and (3) in
an office with at least 20 CPAs or CPA-eligible employees. Demographic information
about the participants is displayed in Table 1. Participants report an average age of 35
years, and 105 months (8.7 years) of experience. Forty-six (61 percent) participants are
female. Participants also reported education and experience common to professionals
working as tax professionals. Specifically, 16 percent work at Big 4 firms, seven percent
at international firms, 16 percent at national firms, 27 percent at regional firms, and 35
percent at local firms. Also, 68 percent of participants report they have researched hobby
loss issues (the issue in the experimental case) in the past. In addition, 71 participants
report either a bachelor’s or master’s degree in business, 55 are CPAs, six are either CIAs
or CMAs, and two are EAs. Demographic characteristics are not significantly different
21
across conditions (all p > 0.23) and do not significantly influence results (all p > 0.30).10
Thus, I conclude random assignment was achieved.
3.2 Design and Manipulations
The hypotheses are tested with a 3 x 1 between-participant experiment.
Participants determine whether Jim and Jennifer Roe can deduct expenses related to
Jennifer’s competitive bicycling activity as a for-profit business (rather than as a hobby),
which would allow the Roes to deduct a loss in the current year.11 Participants are
provided with a draft file memorandum and links to excerpts from IRC §183 and IRS
Regulation 1.183-2 for their reference. I manipulate role by randomly assigning
participants to act as a preparer completing a memorandum or as a reviewer evaluating a
memorandum. Random assignment controls for differences in experience or rank, which
have both been shown to influence advocacy attitudes (Spilker et al. 2016; Bobek et al.
2010).12
In addition to being influenced by the role they occupy, reviewers who view a
preparer’s conclusion in support of the client’s preference may perceive the
recommendation as additional information supporting the client’s preference. Such
additional information may lead to approval bias, which occurs when endorsing another’s
10 All p-values are two-tailed unless noted otherwise. 11 I.R.C. §183 requires taxpayers who engage in activities that might be viewed as hobbies to satisfy a “facts and circumstances” test based on nine identified factors (explained in IRS Regulation 1.183-2) to determine whether the activity is engaged in for profit (business treatment) or is a hobby. The factors are not explicitly weighted or ranked and there are few obvious trends in how the courts have applied the factors. If an activity does not meet the requirements of Reg. 1.183-2, the activity is considered a hobby and the expenses are only deductible as itemized deductions (subject to a floor) and the amount deducted is limited to the amount of income associated with the activity (i.e., no loss can be deducted); taxpayers prefer to treat activities as businesses to allow for the deduction of losses. The rules that apply beginning in 2018 make hobby treatment appear even less favorable because the deductibility of any expenses (rather than just expenses exceeding income) associated with hobbies is unclear. 12 Regression analysis indicates participants in this study exhibited similar trends, such that professionals’ client advocacy attitudes increased with higher rank (b = 0.187, p = 0.108) and increased experience (b = 0.176, p = 0.13).
22
judgment magnifies cognitive biases (Bazerman et al. 2002). To control for the potential
of approval bias, the experiment includes two reviewer conditions manipulating whether
the memo includes a preparer conclusion of business treatment or an explicitly omitted
preparer conclusion.13
3.3 Task and Procedure
The scenario is modified from Bobek et al. (2010), which was designed to be
ambiguous regarding whether the bicycling activity should be viewed as a hobby or
business. The scenario includes several factors that indicate business treatment
(businesslike records and likelihood of future profits) and only one that clearly indicates
hobby treatment (the taxpayer derives significant pleasure from the activity). The
magnitude of the losses in the present study is larger than that in Bobek et al. (2010),
which increases the likelihood the decision appears nontrivial to the Roes’ overall tax
liability. In addition, the Roes are presented as important clients with no clear indications
of high risk because advocacy bias is most likely to occur when client importance is high
and risk is low (Bobek et al. 2010; Kadous et al. 2008). Because the present study also
examines accountability, participants view instructions from the engagement partner that
emphasize the importance of the client and clearly state (1) the partner does not have a
specific outcome in mind and (2) the client prefers business treatment. Specifically, the
13 In practice, a reviewer is unlikely to review a memorandum with no preparer conclusion; however, I include the condition for experimental control. Approval bias suggests a preparer conclusion could exacerbate reviewer advocacy bias if the reviewer perceives the preparer conclusion as additional evidence in support of the client’s preference (i.e., the reviewer may be susceptible to approval bias). Thus, this study isolates the possible effect of approval bias on reviewer judgment by replacing the preparer’s recommendation (i.e., “The Roes’ activity should be treated as a business”) with a note indicating: “For this case study the preparer’s conclusion has been purposely omitted.” Untabulated results indicate the preparer conclusion did not influence (1) reviewers’ likelihood of exhibiting advocacy bias (p = 0.474) or (2) reviewers’ likelihood of assessing factors as supportive of business (p = 0.277). Thus, analysis and hypotheses testing include both reviewer conditions as one collapsed “reviewer” condition because there is no evidence of approval bias.
23
instrument states “Jennifer has become increasingly involved in bicycle racing and would
like to treat her bicycle racing as a business. I have not yet considered the
appropriateness of this decision and need to advise her soon.” This holds constant the
level of outcome accountability felt by participants.
Reviewers are informed they will be reviewing the memo completed by Sam, a
second-year staff at their firm. Preparers are informed the memo is almost complete and
their task is to develop a conclusion regarding the proper tax treatment of the bicycling
activity. The memo is near complete because the study is focused primarily on evidence
evaluation rather than on evidence search. Prior research finds advocacy bias affects both
the search for and evaluation of evidence (Bobek et al. 2010; Kadous et al. 2008; Cloyd
and Spilker 1999). Reviewers usually rely on preparers’ evidence search, but
professionals in both roles evaluate evidence. Thus, participants are presented with
evidence that has already been collected rather than conducting the entire search process,
which maintains similar presentation and cognitive load in all conditions while also
controlling for the effect of preparer stylization on reviewer judgment (Cuccia et al. 2017;
Tan and Trotman 2003).
3.4 Dependent Variable: Evidence Evaluation
The study’s primary analysis examines the relationship between advocacy
attitudes and evidence evaluation. After indicating their conclusion (preparer condition)
or sign-off (reviewer conditions) decisions, participants are asked to indicate whether
each of the nine factors included in Regulation 1.183-2 (see table 3.2) indicate business
or hobby treatment on 6-point scales (where 1 = definitely hobby and 6 = definitely
business). Because Reg. 1.183-2 requires that each determination be based on the facts
24
and circumstances of each case, not all nine factors are relevant to every case. Thus,
participants have the option to indicate that any of the nine factors are not applicable.
The primary dependent variable, FACTORS, is calculated by averaging each
participant’s responses regarding whether each factor indicates business/hobby. The
resulting variable forms a scale ranging from 1 (all factors definitely support hobby) to 6
(all factors definitely support business).14 Advocacy bias is indicated if advocacy
attitudes are related to evidence evaluation. Thus, if participants’ responses to the client
advocacy scale predicts FACTORS (i.e., evidence evaluation), this is evidence of
advocacy bias. As alternative dependent variables, the study also includes measures of
how aggressive the participants’ judgments are. Specifically, I measure participants’
likelihood of recommending business (LIKELY) and assessment of the likelihood the
Tax Court would uphold business treatment (TAXCOURT).
3.5 Accountability
Because tax professionals likely feel accountable to multiple sources, I develop a
scale measuring accountability to the client, firm, and IRS.15 The scale includes 12 items
modified from the scales of Frink and Ferris (1998) and Frink and Klimoski (1998), and
selected items from Donnelly (2017). Specifically, participants respond to items about
perceived accountability, perceived reputation effects, potential future consequences, and
pressure to agree with the client, the firm, and the IRS (See Table 3.3, Panel A for a list
of items). Factor analysis results indicate the items load onto three distinct factors. One
14 I use an average, rather than a sum or factor scores, because factors that are considered “not applicable” are excluded from the calculation of FACTORS for each participant. By averaging responses, the omission of factors for only some participants does not impede comparability. 15 The study’s formal hypotheses focus on accountability to the client and to the firm, but tax professionals may also feel accountable to the IRS. As shown in Table 3, Panel C, accountability to the IRS did not vary by role. However, participants did feel similarly accountable to the IRS as they did to the client and the firm. Accountability to the IRS did not influence their evidence evaluation or conclusions.
25
item, “I felt pressure to make a decision consistent with what the IRS would allow,”
cross-loaded with items related to both the Firm and the IRS. Another item, “I felt
pressure to make a decision consistent with the Roes’ preferences,” cross-loaded with
items related to all three accountability sources (e.g., firm, client, and IRS). Thus, I form
three scales with the remaining items. Accountability to the firm (ACCT_F) includes four
items with a Cronbach’s alpha of 0.775. Accountability to the client (ACCT_C) includes
three items with a Cronbach’s alpha of 0.739. Finally, accountability to the IRS
(ACCT_I) includes three items with a Cronbach’s alpha of 0.752. Mean responses to
each of the three scales are presented in Table 3.3, Panel B and construct correlations are
displayed in Table 3.3, Panel C. Construct correlations (e.g., correlations among the types
of accountability) were all below 0.48, which indicates discriminant validity because
construct correlation is < 0.85 (Klein 2005, 73). All participants, regardless of role, report
feelings of accountability that are not different from the midpoint of the scale (all p >
0.2), indicating participants felt similar levels of moderate accountability to all three
sources.
3.6 Additional Measures
Participants respond to a client advocacy (CA) scale (Bobek et al. 2010; Mason
and Levy 2000), which includes nine items measuring a professional’s attitude towards
his or her attitudes toward the client (See Appendix 1). Next, reviewers (preparers) rate
the hypothetical preparer’s (reviewer’s) competence and client advocacy. Preparers
received very little information about the reviewer, but I include these items to (1) control
26
for any differences in participants’ assumptions about the reviewer’s competence and/or
advocacy and (2) maintain similar time commitment across conditions.16
Next, participants answer three manipulation and attention check items about (1)
the client’s preferred outcome, (2) the role they held during the study (reviewer or
preparer), and (3) which outcome the preparer recommended (reviewers only). Finally,
participants respond to items about the perceived risk preferences of (1) the Roes (the
client) and (2) Oliver and Associates (the firm). Post-experimental questions include
demographic variables and measures of participants’ perceptions of the task. Specifically,
participants indicate which outcome they thought the partner expected, how hard they
worked to complete the task, and how motivated they were to do well. Finally,
participants respond to a scale measuring professional skepticism to control for
differences in participants’ natural levels of skeptical behavior. Bobek, Hageman, and
Radtke (2014) find the five-item questioning mind subscale has the highest correlation
with the full 30-item professional skepticism scale of Hurtt (2010). For brevity, I follow
Bobek et al. (2014) and use the subscale as a measure of professional skepticism.17
16 Preparers’ evidence evaluation (FACTORS) was not significantly influenced by their perceptions of the reviewers’ advocacy attitudes (p = 0.672). Further, time spent on the task did not vary by condition (p = 0.736) or role (p = 0.663). 17 Professional skepticism was not significantly different by ROLE (p = 0.445) and did not influence evaluation of evidence (FACTORS) (p = 0.742). However, professional skepticism did significantly influence likelihood of recommending business treatment (p = 0.054), such that increased professional skepticism led to increased likelihood of business treatment. However, this effect is moderated by role (p = 0.089), such that reviewers with increased professional skepticism were not more likely to indicate business treatment. In addition, professional skepticism is also significantly related to accountability to the client (p = 0.002), firm (p = 0.029), and IRS (p = 0.002), such that increased professional skepticism is related to increased accountability to all three sources. This suggests that the questioning mind construct measured by the professional skepticism subscale may influence both (1) aggressiveness and (2) accountability. Interestingly, this study does not find evidence that professional skepticism influences the effect of advocacy on evidence evaluation (p = 0.209). Future research should further examine the effect of professional skepticism on both aggressiveness and accountability among tax professionals.
27
Table 3.1: Demographic Characteristics
Average Age 35 Average Months of Work Experience 104.91 Gender Male. 37.33% Do not wish to respond. 1.33%
Education Bachelor's in Business (including Accounting or Finance) 78.67% Bachelor's in another area 8.00% Master's in Accounting, Taxation, or Business 41.33% Master’s in Another Area 9.3% Juris Doctorate 2.67% Certifications CPA 73.33% CMA OR CIA 10.67% EA 4.00% None 18.67% How would you classify your firm? Big 4 16.00% International (Other than the Big 4) 6.67% National 16.00% Regional 26.67% Local 34.67%
Current Title Partner/Director. 10.67% Manager/Senior Manager. 44.00% Senior. 26.67% Staff. 18.60%
How familiar are you with Section 183 hobby loss rules? Mean (where 1 = Very Unfamiliar and 7 = Very Familiar) 4.12 (Range 1-6,
Standard Deviation 1.139)
Have you ever researched a hobby loss issue for a client? Yes, many times. 24.00% Yes, once or twice. 44.00% No. 32.00%
Experience Reviewing Others Less than 1 year. 18.67% 1 - 2 years. 24.00% 3 years or more. 50.67% No experience. 6.67%
Currently act as preparer on some engagements Yes. 86.67% No. 13.33%
28
Table 3.2: Reg. 1.183-2 Factors to be considered The table below includes a brief description of each factor, including excerpts from Reg. 1.183-2, and the suggested application of each factor to the Roe’s Case.
Factor and Excerpt from Reg. 1.183-2 Facts of Roes’ Case Manner in which the taxpayer carries on the activity. The fact that the taxpayer carries on the activity in a businesslike manner and maintains complete and accurate books and records may indicate that the activity is engaged in for profit.
Ambiguous: Jennifer keeps track of her earnings and expenses with businesslike records. She has a thorough business plan and plans to continue improving her skills so she can improve her profitability.
The expertise of the taxpayer or his advisors. Preparation for the activity by extensive study of its accepted business, economic, and scientific practices, or consultation with those who are expert therein, may indicate that the taxpayer has a profit motive where the taxpayer carries on the activity in accordance with such practices.
Ambiguous: Jennifer is ranked in two events, which suggests expertise.
The time and effort expended by the taxpayer in carrying on the activity. The fact that the taxpayer devotes much of his personal time and effort to carrying on an activity, particularly if the activity does not have substantial personal or recreational aspects, may indicate an intention to derive a profit.
Ambiguous; Slightly Suggests Hobby: There is no set amount of time required to satisfy this factor; however, Jennifer devotes a significant amount of time to the activity. She trains daily, which could be viewed as regular exercise but may be viewed as business activity. In addition, she adjusts her work schedule at the hospital during racing season to allow for further dedication to the sport.
Expectation that assets used in activity may appreciate in value. The term profit encompasses appreciation in the value of assets, such as land, used in the activity.
Not relevant at all: This factor is not applicable to the current activity as Jennifer’s only asset is the bicycle and trailer, which will both depreciate in value.
The success of the taxpayer in carrying on other similar or dissimilar activities. The fact that the taxpayer has engaged in similar activities in the past and converted them from unprofitable to profitable enterprises may indicate that he is engaged in the present activity for profit, even though the activity is presently unprofitable.
Suggests Business: Jim and Jennifer are both successful in their careers. A review of past returns shows that the Roes’ have started businesses before; however, none of those businesses remain in operation.
29
Table 3.2.: Reg. 1.183-2 Factors to be considered (cont.)
The taxpayer’s history of income or losses with respect to the activity. A series of losses during the initial or start-up stage of an activity may not necessarily be an indication that the activity is not engaged in for profit.
Suggests Hobby: Jennifer has consistently incurred losses.
The amount of occasional profits, if any, which are earned. The amount of profits in relation to the amount of losses incurred, and in relation to the amount of the taxpayer's investment and the value of the assets used in the activity, may provide useful criteria in determining the taxpayer's intent.
Suggests Hobby: Jennifer’s racing has generated sponsorships and some winnings. She expects her winnings to become more frequent, which could produce profits in the future.
The financial status of the taxpayer. The fact that the taxpayer does not have substantial income or capital from sources other than the activity may indicate that an activity is engaged in for profit.
Not relevant at all: Jim and Jennifer are already high-income taxpayers; they are not dependent on her racing activity as a source of income.
Elements of personal pleasure or recreation. The presence of personal motives in carrying on of an activity may indicate that the activity is not engaged in for profit, especially where there are recreational or personal elements involved.
Suggests Hobby: Although she maintains businesslike records and expects to eventually make a profit, Jennifer derives significant recreation and physical pleasure from her bicycling.
30
Table 3.3: Accountability
Panel A: Factor Analysis of Accountability Items
Component
Firm IRS Client 1. I felt pressure to make a decision consistent with what
the partner expected. 0.866 -0.257 0.037
2. I could face consequences if my conclusion did not agree with what the partner expected.
0.849 0.019 -0.073
3. I felt accountable to Oliver and Associates for the decision I just made regarding the Roes' activity.
0.695 -0.116 0.169
4. The decision I just made could influence Oliver and Associates' perceptions of my performance.
0.666 0.204 -0.017
5. I felt pressure to make a decision consistent with what the IRS would allow.
0.447 0.419 -0.044
6. The decision I just made could influence the IRS's perceptions of my performance.
-0.204 0.957 0.055
7. I felt accountable to the IRS for the decision I just made regarding the Roes' activity.
-0.106 0.854 0.012
8. I could face consequences if my conclusion did not agree with what the IRS would allow.
0.381 0.582 -0.236
9. I felt accountable to the Roes for the decision I just made regarding their activity.
-0.094 -0.071 0.899
10. The decision I just made could influence the Roes' perceptions of my performance.
0.166 -0.128 0.824
11. I could face consequences if my conclusion did not agree with the Roes' preferences.
-0.037 0.394 0.590
12. I felt pressure to make a decision consistent with the Roes' preferences.
0.314 0.266 0.392
This panel includes the pattern matrix resulting from Principal Component Analyses with Promax rotation. Items 5 and 12 did not load onto a single factor and are omitted from the resulting scales. Items 1-4 are averaged to create “Accountability to the Firm,” (Cronbach’s Alpha = 0.775); items 6-8 are averaged to create “Accountability to the IRS” (Cronbach’s Alpha =0.752); and items 9-11 are averaged to create “Accountability to the Client” (Cronbach’s Alpha = 0.739).
31
Table 3.3: Accountability (cont.) Panel B: Mean Accountability Responses by Condition (n=75) Accountability Scale Item Preparer Reviewer Total Accountability to the Client (ACCT_C) 4.20 4.09 4.12 Accountability to the Firm (ACCT_F) 3.89 3.97 3.95 Accountability to the IRS (ACCT_I) 4.02 3.96 3.98 This panel includes mean responses to each subscale related to each accountable source, as described in Panel A. Panel C: Pearson Correlations of each Accountability Type (p-values) (n=75)
ACCT_I ACCT_C ACCT_F 0.475 (0.000) 0.387 (0.001) ACCT_I 0.245 (0.034) This panel includes Pearson Correlations of the three types of accountability measures. Construct correlation between constructs (e.g., the types of accountability) were all below 0.48 (Table 3, Panel C), which indicates discriminant validity because construct correlation is <0.85 (Klein 2005, 73).
32
CHAPTER 4
RESULTS
4.1 Manipulation Checks
This study manipulates role by assigning participants to act as either a preparer or
reviewer on a tax research task. To confirm this manipulation, participants are asked
which role they were assigned in the study. All seventy-five participants correctly
responded to the manipulation check with the appropriate role. The study also
manipulates the presence of a preparer conclusion, which is confirmed by asking
participants in the reviewer conditions which preparer conclusion was included in the
memo. Participants assigned to the reviewer_business condition are expected to respond
“treatment as a for-profit business.” 84 percent (21/25) of participants assigned to
reviewer_business correctly responded. In contrast, participants assigned to the
reviewer_omitted condition are expected to respond “the memo did not include a
conclusion.” Surprisingly, 12 (one) of these participants incorrectly responded that the
preparer concluded business (hobby) treatment. Thus, only 48 percent of participants
correctly responded. Overall, 66 percent passed the manipulation check. This suggests
participants may have inferred a conclusion based on the information provided despite its
purposeful omission. Results are not affected by whether the 17 participants who
incorrectly responded to the preparer conclusion manipulation check are included in
analysis. Thus, analyses reported include all 75 participants.
33
4.2 Analysis of Hypothesis 1
Hypothesis 1 predicts professionals occupying the reviewer role are less likely to
exhibit advocacy bias. To test H1, I first establish that participants exhibit advocacy bias
in their decisions which is indicated by a relationship between client advocacy (CA) and
evidence evaluation (FACTORS). Descriptive statistics of participants’ CA, FACTORS,
LIKELY, and TAXCOURT are included in Table 4.1, Panel A. None of the above-
mentioned variables are significantly different by role; however, results do indicate
advocacy bias. As shown in Table 4.1, Panel B, CA significantly influences FACTORS
(p = 0.03).18 In addition, the effect of CA on FACTORS is moderated by ROLE (p =
0.052). Additional untabulated analysis indicates the effect of CA on FACTORS is only
significant in the preparer condition, such that reviewers’ evidence evaluation is not
influenced by advocacy attitudes (p = 0.823). As illustrated in Figure 4.1, this finding
supports H1, providing initial evidence that occupying the reviewer role mitigates
advocacy bias.
4.3 Analysis of Hypothesis 2
Hypothesis two predicts occupying the reviewer role increases professionals’
accountability to the firm. I test H2 with a one-way ANOVA calculated on ACCT_F by
ROLE. As shown in Table 4.2, Panel A, results do not indicate ACCT_F is significantly
influenced by ROLE (p = 0.764).19 Thus, I conclude role did not influence perceived
accountability to the firm.
18 Due to low variance, results are not significant when using a continuous measure of CA. Thus, I perform a median split to classify participants into two groups where “1” indicates high advocacy attitudes (mean CA = 5.3025, std dev. = 0.42713) and “0” indicates low advocacy attitudes (mean CA = 4.0427, std dev. = 0.56519). All reported analyses use this dichotomous variable. 19 This is surprising, given the increased expectation of providing feedback identified by Barrick, et al. (2004); however, this result may be partly driven by the preparers’ unexpected perception of needing to provide “feedback to others on the engagement team regarding their work.” Additional analysis indicates
34
4.4 Analysis of Research Question
The research question investigates whether occupying the reviewer role
influences professionals’ accountability to the client. To test the RQ, I use a one-way
ANOVA calculated on ACCT_C by ROLE. Results indicate ROLE does not influence
ACCT_C (p = 0.634, Table 4.2, Panel A). Thus, I do not find evidence that occupying the
reviewer role affects professionals’ accountability to the client.
4.5 Analysis of H3a: Accountability to the Firm
H3a predicts ACCT_F reduces the likelihood of advocacy bias (the relationship
between CA and FACTORS). To test H3a, I regress ACCT_F and CA on FACTORS to
examine whether ACCT_F decreases the likelihood CA influences evidence evaluation.
Specifically, I examine whether the interaction of ACCT_F and CA is a significant
predictor of FACTORS in equation 1 below.
FACTORS = β1 ACCT_F + β2 CA + β3 ACCT_F * CA. (1) Regression results support a relationship between accountability and advocacy
bias if β3 is significant. As shown in Table 4.2, Panel B, the interaction between ACCT_F
and CA is not significant (p = 0.254), which indicates ACCT_F does not influence the
likelihood of advocacy bias. Thus, H3a is not supported and I find no evidence that
accountability to the firm, as measured in this study, influences the likelihood of
advocacy bias.
4.6 Analysis of H3b: Accountability to the Client
H3b predicts ACCT_C increases the likelihood of advocacy bias. To test H3b, I
regress ACCT_C and CA on FACTORS in equation 2 below to examine whether
participants’ perceptions of the importance of a feedback expectation did not differ by role (p = 0.505, untabulated).
35
accountability to the client increases the likelihood that advocacy attitudes influence
evidence evaluation.
FACTORS = β1 ACCT_C + β2 CA + β3 ACCT_C*CA. (2) Regression results support a relationship between accountability and advocacy
bias if β3 is significant. As shown in Table 4.2, Panel C, ACCT_C significantly influences
the effect of CA on FACTORS (b = -0.399, p = 0.064); thus, ACCT_C appears to
mitigate bias in the evaluation of evidence. Although an interesting finding, this does not
support H3b because H3b predicted ACCT_C would increase advocacy bias and findings
indicate a decrease in advocacy bias. Regardless, this finding suggests accountability to
the client leads professionals to be more objective and provides an unexpected initial
insight into the effect of accountability on tax professional advocacy bias.
4.7 The Interaction of Accountability and Role
Although not formally predicted, I expect accountability and role to jointly
influence the effects of advocacy attitudes on evidence evaluation (See Figure 2.1). H1
examines the effect of ROLE on the likelihood of advocacy bias; H2 examines the effect
of ROLE on accountability; and H3a and H3b examine the effect of accountability to the
firm and accountability to the client, respectively, on the likelihood of advocacy bias. As
supplemental analysis, I also examine the effects of ROLE and ACCT_F/ACCT_C on
likelihood of advocacy bias in a more comprehensive test of the model. Specifically, I
regress ACCT_F (ACCT_C), ROLE, and CA on FACTORS as shown in equations 3 (4),
FACTORS = β1 ACCT_C + β2 CA + β3 Role + β4 ACCT_C*CA (4) + β5 ACCT_C*Role + β6 CA*Role + β7 ACCT_C*CA*Role. As shown in Table 4.2, Panel D, I do not find evidence that ACCT_F influences
the likelihood of advocacy bias (p = 0.505). Given the findings in H1 that the reviewer
role reduces the likelihood of advocacy bias, I also test for a moderating effect of role. I
do not find that ROLE influences the relationship between ACCT_F and FACTORS.
However, as shown in Table 4.2, Panel E, accountability to the client presents some
interesting findings. Specifically, I find a significant interaction between CA and
ACCT_C (p = 0.024), which supports the findings in H3b that ACCT_C reduces the
likelihood that CA influences FACTORS. Thus, accountability to the client moderates
the effect of advocacy attitudes on evidence evaluation.
4.6 Supplemental Analysis: Aggressiveness
I also examine whether participants’ advocacy attitudes (CA), potential advocacy
bias (CA * FACTORS), and role (ROLE) influence the aggressiveness of final
recommendations (LIKELY) (see equation 5 below).
LIKELY = β1 FACTORS + β2 CA + β3 Role + β4 FACTORS*CA (5) + β5 FACTORS*Role + β6 CA*Role + β7 FACTORS*CA*Role. As discussed earlier, advocacy bias occurs during the evidence evaluation
process. However, it can also influence the aggressiveness of professionals’ judgments.
As shown in Table 4.4, Panel A, I find LIKELY is significantly influenced by (1) ROLE
(p = 0.042), (2) CA (p < 0.001), and (3) the interaction of CA and ROLE (p < 0.001). I do
not find a significant three-way interaction between CA, ROLE, and FACTORS (p =
0.553). Thus, results indicate higher client advocacy attitudes (1) increase the likelihood a
professional’s conclusion/sign-off decision (LIKELY) will support business treatment
37
and (2) decreases the effect of ROLE on conclusion/sign-off decision (LIKELY). Further
regression analysis, split by role and displayed in Table 4.3, Panel B, finds LIKELY is
influenced primarily by CA among preparers (p < 0.001) and by FACTORS among
reviewers (p = 0.001). Thus, professionals acting as reviewers are more focused on the
evidence evaluation and those acting as preparers are more focused on their own
advocacy attitudes. This provides further evidence that occupying the reviewer role
changes how professionals approach a task, which can lead to increased objectivity (i.e.,
less bias).
4.8 Supplemental Analysis: Partner Preference
The study explicitly states the partner has “not yet considered the appropriateness
of this decision,” to avoid participants’ attitude shifting toward an accountability source’s
known outcome preference. Such attitude shifting could lead participants to
unintentionally align their evaluation to agree with the partner’s preference. However, as
shown in Table 4.4, when asked what they thought the partner expected, only 17 (23
percent) participants reported that the partner did not have a preference. Of the other 58
participants, 46 (12) assumed the partner expected a business (hobby) recommendation.
These assumptions did not vary across conditions (p = 0.878) and did not influence
FACTORS (p = 0.238).
However, the perceived partner expectation does influence LIKELY (p = 0.017).
Specifically, participants who assume the partner preferred business treatment are more
likely to recommend business treatment than those who assume the partner preferred
hobby and those who assume the partner does not have a preference. Interestingly, the
effect of perceived partner preference on LIKELY is moderated by ROLE (p = 0.018)
38
such that perceived partner preference has a significant effect on LIKELY only for
preparers.
In addition, when comparing participants who correctly answered that the partner
did not have a preference to those who indicate the partner did have a preference, those
who perceived a preference are more likely to both recommend business treatment (p =
0.031) and to evaluate factors as supportive of business treatment (p = 0.095). Given the
influence of assuming a partner preference, I also examine the effect of assuming
business (compared to assuming hobby). Results are shown in Table 7, Panel D. I find
participants who assume the partner preferred business treatment are significantly more
likely to recommend business treatment than those who assume the partner preferred
hobby treatment (p = 0.053) but are not more likely to evaluate the factors as supportive
of business treatment (p = 0.740). This suggests the perceived partner preference for
business treatment influence how participants integrated their evidence evaluation into
their conclusions. Specifically, participants who assume the partner had a preference for
business exhibit bias in the development of their conclusions.
4.9 Supplemental Analysis: Factors Included in Conclusions
As an additional test of whether the reviewer role affects how professionals
evaluate evidence, I also examine (1) which factors are predictive of LIKELY and (2)
whether different factors were more influential by role. To examine this, I conduct a
stepwise regression, as shown below in equation 6, on three groups: (1) the entire sample
(n=75), (2) only reviewers (n=50), and (3) only preparers (n=25).
LIKELY = β1 FACTOR1 … + … β9 FACTOR9 (6)
39
Results of the three regressions are presented in Table 4.5. For the entire sample, I
find professionals’ likelihood of recommending (preparers) or signing off (reviewers) on
business treatment is significantly influenced by “elements of personal pleasure” (p =
0.001), “success in prior ventures” (p = 0.006), and “occasional profits” (p = 0.017). This
suggests professionals’ conclusions are formed based on their evaluation of these three
factors.
However, regression results indicate reviewers focus on different factors than
preparers when developing conclusions. Specifically, reviewers focused only on
[1.476] This panel includes the mean responses [standard deviations] related to CLIENT ADVOCACY (mean response to nine items on the Client Advocacy scale of Bobek et al. (2010)), FACTORS (mean response of nine factors from Reg. Sec. 1.183-2), LIKELY (the likelihood of concluding or signing off on a conclusion of business), and TAXCOURT (the likelihood the Tax Court would uphold a conclusion of business treatment) by ROLE and in total. Panel B: Effect of Client Advocacy and Role on Evaluation of Factors (n=75)
Source df Mean Square F-statistic p-value Role 1 0.300 0.408 0.525 Client Advocacy 1 3.596 4.891 0.030 Role * Client Advocacy 1 2.872 3.906 0.052 This panel includes the ANOVA results calculating the effect of ROLE and CLIENT ADVOCACY on evaluation of FACTORS. ROLE is coded such that “1” signifies reviewer and “0” signifies preparer. CLIENT ADVOCACY is the average of the nine-item scale of Bobek et al. (2013), dichotomized by a median split such that “1” indicates high CLIENT ADVOCACY and “0” indicates low CLIENT ADVOCACY.
41
Table 4.2: Analysis of Accountability Panel A: Effect of Role on Accountability to the Client, Firm, and IRS (H2 and RQ, n=75) Source df Mean
Square F-statistic p-value
Accountability to the Client (ACCT_C) 1 0.190 0.229 0.634 Accountability to the Firm (ACCT_F) 1 0.060 0.091 0.764 Accountability to the IRS (ACCT_I) 1 0.107 0.102 0.750 This panel includes the results from three independent univariate tests of the effect of the reviewer role on accountability to (1) the client, (2) the firm, and (3) the IRS. Results indicate the reviewer role is not significantly related to any of the three accountability measures (all p>0.5). Panel B: Effect of ACCT_F on the relationship between CLIENT ADVOCACY and FACTORS (H3a, n=75) B Std. Error t-statistic p-value CLIENT ADVOCACY 0.364 0.200 1.821 0.073 ACCT_F -0.077 0.158 -0.491 0.625 CLIENT ADVOCACY * ACCT_F -0.239 0.208 -1.149 0.254
This panel includes regression results testing the effect of CLIENT ADVOCACY and ACCT_F on FACTORS. CLIENT ADVOCACY is calculated as described in Table 4.1, Panel B. ACCT_F is calculated as described in Table 3, Panel A. FACTORS is the mean of each participants’ response related to whether each of the nine factors in Reg. Sec. 1.183-2 indicated hobby or business treatment such that “1” indicates definitely hobby and “6” indicates definitely business. Panel C: Effect of ACCT_C on the relationship between CLIENT ADVOCACY and FACTORS (H3b, n=75) B Std. Error t-statistic p-value CLIENT ADVOCACY 0.389 0.201 1.935 0.057 ACCT_C 0.038 0.138 0.275 0.784 CLIENT ADVOCACY * ACCT_C -0.399 0.212 -1.885 0.064 This panel includes regression results testing the effect of CLIENT ADVOCACY and ACCT_C on FACTORS. CLIENT ADVOCACY is calculated as described in Table 4.1, Panel B. ACCT_C is calculated as described in Table 3, Panel A. FACTORS is calculated as described above in Panel A.
42
Table 4.2: Analysis of Accountability (cont.) Panel D: Effect of ACCT_F and ROLE on the relationship between CLIENT ADVOCACY and FACTORS (n=75) B Std. Error t-statistic p-value CLIENT ADVOCACY 1.077 0.349 3.082 0.003 ROLE 0.563 0.282 1.997 0.050 ACCT_F -0.187 0.279 -0.670 0.505 CLIENT ADVOCACY * ROLE -1.020 0.423 -2.412 0.019 CLIENT ADVOCACY * ACCT_F -0.662 0.432 -1.532 0.130 ROLE * ACCT_F 0.125 0.271 0.464 0.644 CLIENT ADVOCACY * ROLE * ACCT_F
0.484 0.491 0.986 0.328
This panel includes regression results testing the effect of CLIENT ADVOCACY, ROLE, and ACCT_F on FACTORS. CLIENT ADVOCACY is calculated as described in Table 4.1, Panel B. ROLE is a categorical variable such at “0” indicates preparer and “1” indicates reviewer. ACCT_F is calculated as described in Table 3, Panel A. FACTORS is calculated as described in Table 4.2, Panel B. Panel E: Effect of ACCT_C and ROLE on the relationship between CLIENT ADVOCACY and FACTORS (n=75) B Std. Error t-statistic p-value CLIENT ADVOCACY 0.970 0.333 2.910 0.005 ROLE 0.588 0.278 2.114 0.038 ACCT_C -0.068 0.224 -0.304 0.762 CLIENT ADVOCACY * ROLE -0.927 0.410 -2.260 0.027 CLIENT ADVOCACY * ACCT_C -0.920 0.398 -2.312 0.024 ROLE * ACCT_C 0.208 0.253 0.823 0.413 CLIENT ADVOCACY * ROLE * ACCT_C
0.604 0.466 1.297 0.199
This panel includes regression results testing the effect of CLIENT ADVOCACY, ROLE, and ACCT_C on FACTORS. CLIENT ADVOCACY is calculated as described above in Table 4.1, panel B. ROLE is a categorical variable such that “0” indicates preparer and “1” indicates reviewer. ACCT_C is calculated as described in Table 3, Panel A. FACTORS is calculated as described above in Table 4.2, Panel B.
43
Table 4.3: Supplemental Analysis of Aggressiveness
Panel A: Effect of CA, FACTORS, and ROLE on Conclusion Aggressiveness (LIKELY) (n=75)
This panel includes regression results testing the effect of CLIENT ADVOCACY, FACTORS, and ROLE on LIKELY. CLIENT ADVOCACY is calculated as described in Table 4.1, Panel B. FACTORS are calculated as described in Table 4.2, Panel B. ROLE is described in Table 4.2, Panel B. Panel B: Effect of CA and FACTORS on Conclusion Aggressiveness (LIKELY) by ROLE Preparers (n=25) Reviewers (n=50)
This panel includes regression results testing the effects of CLIENT ADVOCACY and FACTORS on LIKELY by ROLE. CLIENT ADVOCACY is calculated as described in Table 4.1, Panel B. FACTORS are calculated as described in Table 4.2, Panel A. LIEKLY is the participants’ likelihood of either concluding or signing off on a conclusion that the activity should be treated as a business.
Which of the following outcomes do you think the engagement partner expected?
Total
Treatment as a for-profit business.
Treatment as a hobby.
The partner didn't have an
expectation. ROLE Preparer 15 6 4 25
Reviewer 31 6 13 50 Total 46 12 17 75 Participants responded to the question “Which of the following outcomes do you think the engagement partner expected?” with three possible responses. All participants were expected to indicate that the partner didn’t have an expectation; however, as shown above, many participants imputed a partner preference. Panel B. ANOVA
Source df Mean Square F-statistic p-value PERCEIVED PARTNER PREFERENCE
1 0.300 0.408 0.525
ROLE 1 3.596 4.891 0.030 PERCEIVED PARTNER PREFERENCE * ROLE
1 2.872 3.906 0.052
This panel includes the ANOVA results calculating the effect of the reviewer role and perceived partner preference on LIKELY. ROLE is coded such that “1” signifies reviewer and “0” signifies preparer. Perceived partner preference is also an indicator variable, such that “1” is treatment as a for-profit business, “2” is treatment as a hobby, and “3” is the partner didn’t have an expectation. Panel C. ANOVA examining the effect of perceiving a partner preference
Source df Mean Square
F Sig.
LIKELY 1 6.419 4.845 .031 FACTORS 1 2.172 2.860 .095 This panel includes the ANOVA results calculating the effect of perceiving a partner preference (either hobby or business) on (1) the likelihood of recommending or signing off on business treatment and (2) the evaluation of factors.
45
Table 4.4 Supplemental Analysis of Partner Preference (cont.) Panel D. ANOVA examining the effect of perceiving a partner preference for business (n=58)
Source df Mean Square F Sig. LIKELY 1 4.606 3.894 0.053 FACTORS 1 0.080 0.112 0.740
This panel includes the ANOVA results calculating the effect of perceiving a partner preference of business (compared to hobby) on (1) the likelihood of recommending or signing off on business treatment and (2) the evaluation of factors.
46
Table 4.5 Supplemental Analysis of the Influence of FACTORS on LIKELY (n=75) Panel A: Stepwise Regression of Individual FACTORS on LIKELY (n=75) Model Unstandardized
This panel includes stepwise regression results regressing the nine individual factors used to calculate FACTORS on LIKELY. Stepwise criteria included p <=0.05 for entry and p >= 0.10 for removal. Panel B: Stepwise Regression of Individual FACTORS on LIKELY for Reviewers (n=25) Model Unstandardized
Coefficients Standardized Coefficients
t Sig.
B Std. Error Beta 1 F_OCCPROFITS 0.538 0.121 0.666 4.463 <0.001 This panel includes stepwise regression results regressing the nine individual factors used to calculate FACTORS on LIKELY using only the reviewer condition. Stepwise criteria included p <=0.05 for entry and p >= 0.10 for removal. Panel C: Stepwise Regression of Individual FACTORS on LIKELY for Preparers (n=50) Model Unstandardized
This panel includes stepwise regression results regressing the nine individual factors used to calculate FACTORS on LIKELY for those in the preparer condition. Stepwise criteria included p <=0.05 for entry and p >= 0.10 for removal.
47
Figure 4.1. This figure shows the effects of client advocacy (CA) and role (ROLE) on evidence evaluation (FACTORS). For this figure, CA is a categorical variable based on the dichotomized responses to the client advocacy scale where “1” indicates high CA (above the median) and “0” indicates low CA (below the median). FACTORS is the mean evaluation of nine individual factors from Reg. 1.183-2. The interaction between role and client advocacy is significant (p = 0.026, one-tailed).
48
CHAPTER 5
CONCLUSION
This study examines whether the review process mitigates advocacy bias by
isolating the effect of the reviewer role on tax professional judgments. Specifically, I
randomly assign professionals to the preparer and reviewer roles, which controls for
individual differences in experience and rank. Prior literature in audit suggests reviewers
(compared to preparers) demonstrate a heightened focus on objectivity (Reimers and
Fennema 1999). I find tax professionals occupying the reviewer role exhibit similar
behavior. Specifically, they are less likely to exhibit advocacy bias than those in the
preparer role, focus on different types of evidence than preparers, and are less influenced
by a perceived partner preference. The reviewer role does not appear to prompt an
increase in perceived accountability; however, this study provides initial evidence that
increased accountability influences how preparers evaluate evidence. In addition,
supplemental analysis examines how advocacy attitudes, role, and evidence evaluation
influence professionals’ conclusions. Results indicate preparers’ conclusions are
significantly influenced by advocacy attitudes whereas reviewers’ conclusions are
significantly influenced by evidence evaluation. This suggests occupying the reviewer
role prompts professionals not only to objectively evaluate evidence, as found in H1, but
also to objectively determine a conclusion. Thus, this study provides evidence on many
differences in decision making among professionals in the reviewer role when compared
to those in the preparer role, including a reduced influence of advocacy bias on judgment.
49
The study is subject to several limitations. First, the conclusions are based on
hypothetical judgments using limited information to minimize the time commitment
required of tax professionals who were willing to participate. Future research could
examine whether the effect of the reviewer role holds in other contexts. Second, the
scenario used in this study was related to an ambiguous issue within an individual tax
compliance context. The findings may not translate directly to tax professional judgments
in more complex business tax compliance issues. Finally, the accountability measure
included in this study was based on selected measures identified in prior accountability
research. Given the limited variation in these measures, future studies could further
develop measures of accountability to multiple sources in both tax and audit contexts.
This study contributes to the literature in several ways. First, by isolating the
effect of the reviewer role on tax professional judgments, the findings add to a growing
literature (e.g., Cuccia et al. 2017; Barrick et al. 2004) studying whether the review
process mitigates advocacy bias. Specifically, this study establishes the reviewer role as a
moderator of the effect of advocacy attitudes on evidence evaluation. Although prior
studies have separately examined advocacy bias in both preparer and reviewer judgments,
those studies have primarily focused on (1) whether reviewers detect bias through
specific documentation cues or writing styles or (2) whether professionals agreed with the
client-preferred, tax minimizing position. Rather than focusing on whether reviewers
recognize bias in subordinate recommendations or agree with aggressive conclusions, this
study directly compares the evidence evaluation processes of professionals who have
been assigned to be reviewers or preparers. I find reviewers are less likely to exhibit
advocacy bias in their own evidence evaluation, which suggests future research could
50
examine why reviewers have struggled to detect bias in subordinate work (Cuccia, et al.
2017; Barrick et al. 2004; Hatfield et al. 2000).
Second, this study finds professionals may assume that engagement partners
prefer to support client preferences. Specifically, the experimental materials explicitly
instructed participants that the partner had not considered the issue and did not have an
opinion; however, the majority of the participants reported that they thought the partner
preferred supporting the client preference. Further, this perceived preference significantly
influenced likelihood of supporting the client’s preference of business treatment. Thus,
professionals may engage in attitude shifting to meet their inappropriately inferred
partner preference even when the partner attempts to prevent such a response. Hence,
firms may need to consider more strongly emphasizing the importance of independent
judgment.
Third, this study highlights several ways in which professionals occupying the
reviewer role process information and make decisions differently than those in a preparer
role. In addition to the reduced influence of advocacy bias, I also find reviewers are less
influenced by the partner’s preference for supporting the client. Although I intended to
hold the perceived partner’s preference constant, the participants’ assumption that the
preference was to support the client’s preference provided an opportunity to examine the
effect of such an assumption. Findings suggest reviewers may be less susceptible to
attitude shifting, which may in turn reduce the likelihood of predecisional distortion
(Wilkes 2002). In addition, I find reviewers and preparers focus on different pieces of
evidence when making their conclusions. Specifically, reviewers focus on objective
evidence (e.g., whether the client’s activity has produced profits) and preparers focus on
51
evidence that is more subjective (e.g., for example, the elements of personal pleasure
experienced as part of the activity). This finding provides additional evidence that
reviewers are more objective, and less biased, than preparers.
Fourth, the accountability measure used in this study examines the effects of
accountability to the firm and accountability to the client. Specifically, accountability to
the client appears to influence the likelihood of advocacy bias but accountability to the
firm does not. By studying accountability in a tax context, the present study examines the
effect of accountability to multiple sources on professional judgment in an environment
where professionals are likely to feel strong accountability to the client, for whom they
are required to advocate, and to their employer. This potentially contributes to the audit
literature because prior literature has established (1) that accountability influences
multiple auditor judgments and (2) auditors face multiple accountability sources. Thus,
future studies could further investigate the competing effects of accountability to multiple
sources on auditor judgments in both auditor-client and auditor-auditor interactions.
Finally, this study provides initial evidence that firms may be able to decrease the
likelihood of advocacy bias at the preparer level by focusing on accountability to the
client. Although unexpected a priori, this finding provides an interesting insight into the
relationship between preparers and their clients that can be further explored in future
research.
This study expected the effect of role on tax professional judgment to be
explained by differences in accountability. Although results do not support that
expectation, results do provide robust evidence that the judgment and decision making
processes of reviewers are quite different than those of preparers. The study’s findings
52
could be explained by an increase in skepticism prompted by occupying the reviewer
role. I measure a subset of the trait of professional skepticism and find no differences;
however, it is possible state skepticism, or a more general construct of dubiousness, may
explain differences in judgment and decision making processes across roles. Specifically,
reviewers are entering the decision making process after an initial decision has already
been made. Thus, rather than integrating all of the evidence to reach a conclusion,
reviewers are able to focus on inconsistencies and alternative explanations. In addition,
evidence that seemed irrelevant to a preparer may appear more salient to a reviewer if it
does not agree with the preparer’s conclusion. Future research could examine whether
occupying the reviewer role prompts an increase in state skepticism.
Alternatively, the differences identified in this study could also be explained by
the attitude change literature (Chaiken 1980), which suggests individuals who are more
highly involved in a decision (e.g., a preparer forming a conclusion) are more focused on
specific arguments and a conclusion whereas those who are less involved (e.g., a
reviewer evaluating an existing conclusion) are more attentive to cues outside the specific
argument and conclusion (e.g., inconsistent information or surprising evidence).
Similarly, it is also possible preparers feel further separated, or more insulated, from the
final decision maker. Thus, they may feel more permission to be incorrect or to rely on
the review process to detect issues in judgment (Westermann, Bedard, and Earley 2015).
Another possible explanation is impression management. Preparers may feel more
pressure to motivate a conclusion in alignment with the client’s preference with the
knowledge that a reviewer will curb unreasonable aggressiveness. Future evidence can
53
examine these and other social psychology theories to further explain the effects of
occupying the reviewer role.
54
REFERENCES
Agoglia, C. P., T. Kida, and D. M. Hanno. 2013. The effects of alternative justification memos on the judgments of audit reviewees and reviewers. Journal of Accounting Research, 41: 33–46.
AICPA. 2006. AU Section 230 Due Professional Care in the Performance of Work. ———. 2010. Statements on Standards for Tax Services, 1-7. New York, NY. ______. 2016. Positions in Public Accounting. Retrieved from
Ammeter, A. P., Douglas, C., Ferris, G. R., & Goka, H. 2004. A social relationship
conceptualization of trust and accountability in organizations. Human Resource Management Review, 14(1), 47-65.
Asare, S. K., G. M. Trompeter, and A. M. Wright. 2000. The effect of accountability and
time budgets on auditors’ testing strategies. Contemporary Accounting Review 17 (4): 539-560.
Barrick, J. A., C. B. Cloyd, and B. C. Spilker. 2004. The influence of biased tax research
memoranda on supervisors’ initial judgments in the review process. The Journal of the American Taxation Association 26 (1): 1–19.
Bazerman, M. H., Loewenstein, G., & Moore, D. A. 2002. Why good accountants do bad
audits. Harvard Business Review 80 (11): 96-103. Biddle, B. J. 1979. Role theory: expectations, identities, and behaviors. New York:
Academic. Bobek, D. D., A. M. Hageman, and R. C. Hatfield. 2010. The role of client advocacy in
the development of tax professionals’ advice. The Journal of the American Taxation Association 32 (1): 25–51.
Bobek, D. D., A. M. Hageman, and R. R. Radtke. 2014. A research note on the
relationship between professional skepticism and client advocacy. Advances in Accounting Behavioral Research 17: 161-182.
55
Bobek, D. D., D. Dalton, A. M. Hageman, and R. R. Radtke. 2016. An experiential investigation of tax professionals contentious interactions with clients. The Journal of the American Taxation Association, forthcoming.
Chaiken, S., 1980. Heuristic versus systematic information processing and the use of
source versus message cues in persuasion. Journal of personality and social psychology, 39(5): 752-766.
Cloyd, C. B., 1997. Performance in tax research tasks: The joint effects of knowledge and
accountability.” The Accounting Review 72 (1): 111 -131. Cloyd, C. B., and B. C. Spilker. 1999. The influence of client preferences on tax
professionals’ search for judicial precedents, subsequent judgments, and recommendations. The Accounting Review 74 (3): 299–322.
Collins, J. H., V. C. Milliron, and D. R. Toy. 1992. Determinants of tax compliance: a
contingency approach. The Journal of the American Taxation Association 14 (2): 1–30.
Cuccia, A. D., A. M. Magro, and A. Whisenhunt. 2017. The Potential for the Review
Process to Detect and Mitigate Advocacy Bias. Working Paper, University of Oklahoma.
Donnelly, A. 2017. Three Studies Examining Accountability in Auditing. Unpublished
Dissertation; University of Central Florida. Fleischman, G. M., and T. Stephenson. 2012. Client variables associated with four key
determinants of demand for tax preparer services: An exploratory study. Accounting Horizons 26 (3): 417–436.
Frink, D. D., and G. R. Ferris. 1998. Accountability, impression management and goal
setting in the performance evaluation process. Human Relations 51 (10): 1259-1283. Frink, D. D., & R. J. Klimoski. 1998. Toward a theory of accountability in organizations
and human resource management. In. G.R. Ferris (Ed.), Research in personnel and human resources management 16: 1-51. Elsevier Science/JAI Press.
Gibbins, M. and J. D. Newton. 1994. An empirical exploration of complex accountability
in public accounting. Journal of Accounting Research 32 (2): 165-186. Gibbins, M. and K. Trotman. 2002. Audit review: Managers’ interpersonal expectations
and conduct of review. Contemporary Accounting Review 19(3): 411-444. Hall, A. T., F.R. Blass, G. R. Ferris, & R. Massengale. (2004). Leader reputation and
accountability in organizations: Implications for dysfunctional leader behavior. The Leadership Quarterly, 15(4): 515-536.
56
Hall, A. T., D. D. Frink, & M. R. Buckley. 2017. An accountability account: A review
and synthesis of the theoretical and empirical research on felt accountability. Journal of Organizational Behavior 38(2): 204-224.
Hatfield, R. C. 2000. The effect of accountability on the evaluation of evidence: A tax
setting. In Advances in Taxation (pp. 105-125). Emerald Group Publishing Limited. Hatfield, R. C. 2001. The effect of staff accountant objectivity in the review and decision
process: a tax setting. The Journal of the American Taxation Association 23 (1): 61–74.
Hoffman, V. B. and J. M. Patton. 1997. Accountability, the dilution effect, and
conservatism in auditors’ fraud judgments. Journal of Accounting Research 35 (2): 227-237.
Hurtt, R.K. 2010. Development of a scale to measure professional skepticism. Auditing:
A Journal of Practice and Theory 29 (1): 149–171. Internal Revenue Service. 2014. Circular 230. Regulations Governing Practice before the
Internal Revenue Service of the Treasury. https://www.irs.gov/tax-professionals/circular-230-tax-professionals
Ismail, Z., and K. T. Trotman. 1995. The impact of the review process in hypothesis
generation tasks. Accounting, Organizations and Society 20 (5): 345–357. Johnson, L. M. 1993. An empirical investigation of the effects of advocacy on preparers’
evaluations of judicial evidence. The Journal of the American Taxation Association 15 (1): 1–22.
Johnson, V. E., and S. E. Kaplan. 1991. Experimental evidence on the effects of
accountability on auditor judgments. Auditing: A Journal of Practice and Theory 10 (Suppl): 96-107.
Kadous, K., A. M. Magro, and B. C. Spilker. 2008. Do effects of client preference on
accounting professionals’ information search and subsequent judgments persist with high practice risk? The Accounting Review 83 (1): 133–156.
Kadous, K., and Magro, A. M. 2001. The effects of exposure to practice risk on tax
professionals' judgements and recommendations. Contemporary Accounting Research, 18 (3): 451-475.
Kahle, J. B., and R. A. White. 2004. Tax professional decision biases: the effects of
initial beliefs and client preference. The Journal of the American Taxation Association 26: 1–29.
57
Klein, R. B. 2005. Principles and Practices of Structural Equation Modeling. Second edition. New York, NY: The Guilford Press.
Koonce, L., U. Anderson, and G. Marchant. 1995. Justification of Decisions in Auditing.
Journal of Accounting Research 33 (2): 369-384. Libby, R., and K. T. Trotman. 1993. The review process as a control for differential recall
of evidence in auditor judgments. Accounting, Organizations and Society 18 (6): 559–574.
Mason, J. D., and L. G. Levy. 2001. The use of the latent constructs method in behavioral
accounting research: The measurement of client advocacy. In Advances in Taxation (pp. 123-139). Emerald Group Publishing Limited.
Pinsker, R., R. Pennington, and J. K. Schafer. 2009. The influence of roles, advocacy,
and adaptation to the accounting decision environment. Behavioral Research in Accounting 21 (2): 91–111.
Pennington, R., Schafer, J. K., and Pinsker, R. 2016. Do auditor advocacy attitudes
Reimers, J. L., and M. G. Fennema. 1999. The audit review process and sensitivity to
information source objectivity. Auditing: A Journal of Practice & Theory 18 (1): 117–123.
Ricchute, D. 1999. The effect of audit seniors’ decisions on working paper
documentation and partners’ decisions. Accounting, Organizations, and Society 24: 155-171.
Roberts, M. L. 1998. Tax accountants’ judgment/decision-making research: a review and
synthesis. The Journal of the American Taxation Association 20 (1): 78–121. Robertson, C. B. 2009. Judgment, identity, and independence. Connecticut Law Review
42 (1): 1-48. Siegel-Jacobs, K., & J. F. Yates. 1996. Effects of procedural and outcome accountability
on judgment quality. Organizational Behavior and Human Decision Processes, 65(1), 1-17.
Spilker, B.C., B. W. Stewart, J. H. Wilde, and D. A. Wood. 2016. A comparison of U.S.
and offshore-Indian tax professionals’ client advocacy attitudes and client recommendations. The Journal of the American Taxation Association 33 (2): 51-66.
58
Tan, H-T and K. T. Trotman. 2002. Reviewers’ Responses to Anticipated Stylization Attempts by Preparers of Audit Workpapers. The Accounting Review 78 (2): 581-604.
Tetlock, P. E. 1983. Accountability and complexity of thought. Journal of Personality
and Social Psychology 45 (1): 74–83. Tetlock, P. E., 1985. Accountability: The neglected social context of judgment and
choice. Research in Organizational Behavior 7(1): 297-332. Tetlock, P. E. 1999. Accountability theory: Mixing properties of human agents with
properties of social systems. Shared cognition in organizations: The management of knowledge: 117-137.
Westermann, K.D., J.C. Bedard, and C. E. Earley. 2015. Learning the “craft” of auditing:
A dynamic view of auditors’ on-the-job learning. Contemporary Accounting Research 32(3): 864-896.
Wilks, T.J. 2002. Predecisional distortion of evidence as a consequence of real-time audit
review. The Accounting Review 77(1): 51-71. Yip-Ow, J. and H-T. Tan. 2000. Effects of the preparer’s justification on the reviewer’s
hypothesis generation and judgment in analytical procedures. Accounting, Organizations, and Society 25: 203-215.
1. Where no judicial authority exists with respect to an issue, I feel that the Roes
are entitled to take the most favorable tax treatment.
2. I feel I should apply ambiguous tax law to the Roes’ benefit.
3. Generally speaking, my loyalties are first to the tax system, then to the Roes.
4. I believe it is important that I encourage the Roes to pay the least amount of taxes possible.
5. In an instance where no judicial authority exists with respect to an issue and
where the Code and Regulations are ambiguous, I feel that the Roes are entitled to take the most favorable tax treatment.
6. When examining a tax return, I tend to point out to the Roes reasonable positions
they could have taken which would have contributed to minimizing their tax liability.
7. I always interpret unclear/ambiguous laws in favor of the Roes.
8. It is important to use trends in the law by trying to establish a pattern of more
favorable treatment for the Roes and then extending this pattern to the Roes’ position.
9. The Roes have the right to structure transactions in ways that yield the best tax
result, even if the law is unclear in an area.
60
Professional Skepticism – Questioning Mind Subscale (where 1 = strongly disagree and 6 = strongly agree) (Hurtt 2010) (Mean 4.592, SD 0.826)
1. I enjoy trying to determine if what I read or hear is true.
2. I usually notice inconsistencies in explanations.
3. My friends tell me that I often question things I see or hear.
4. I frequently question things that I see or hear.
5. I often reject statements unless I have proof that they are true.
61
Felt Accountability Scale (where 1 = strongly disagree and 7 = strongly agree) (Modified from Frink and Farris 1998, Frink and Klimoski 1998, and Donnelly 2017) Please indicate the extent to which you agree with the following statements. (Accountability to the Client: Mean 4.129, SD 0.905)
1. I felt accountable to the Roes for the decision I just made regarding their activity. 2. I felt pressure to make a decision consistent with the Roes’ preferences. *
3. I could face consequences if my conclusion did not agree with the Roes’
preferences.
4. The decision I just made could influence the Roes’ perceptions of my performance.
(Accountability to the IRS: Mean 3.947, SD 1.015)
5. I felt accountable to the IRS for the decision I just made regarding the Roes’ activity.
6. I felt pressure to make a decision consistent with what the IRS would allow. *
7. I could face consequences if my conclusion did not agree with what the IRS would allow.
8. The decision I just made could influence the IRS’s perceptions of my
performance. (Accountability to the Firm: Mean 3.980, SD 0.808)
9. I feel accountable to Oliver and Associates for the decision I just made regarding the Roes' activity.
10. I felt pressure to make a decision consistent with the partner’s expectation.
11. I could face consequences if my conclusion did not agree with the partner’s
expectation.
12. The decision I just made could influence Oliver and Associates’ perceptions of my performance.
*Two items did not load onto a single factor and are omitted from scale means.
62
APPENDIX B
RESEARCH MEMORANDUM PROVIDED TO PARTICIPANTS
To: File
RE: Roe Bicycle Activity (2015-2017)
Date: July 1, 2017
FACTS Jennifer has become increasingly active in off-road bicycle racing. Since the beginning of 2015, she has participated in approximately 60 races. The National Off-Road Bicycle Association has recently ranked Jennifer in the top 100 for both the dual slalom and the downhill races. The most recent rankings are 89th in the dual slalom and 95th in the downhill. These rankings have led to a few sponsorships that have provided Jennifer with clothing, and, occasionally, entry fees. Although Jennifer has won a few of the races she has entered, her annual expenses related to racing (e.g., bicycle repairs, transportation, lodging, and entry fees) exceed her racing income. Jim and Jennifer tell you that she trains daily. During racing season, she only works 30 hours per week (rather than her standard 40) so she can make the most of the racing events. Jennifer has kept businesslike records of all of her earnings and expenses for the last three years (including 2017) and she has a thorough business plan. For 2015 and 2016, her winnings were $990 and $2,180, respectively. Her expenses for those years were $15,200 and $21,700, respectively. With the 2017 season half-way over, she estimates that her earnings will be around $4,000 while her expenses will be around $20,000. Cash prizes run as high as $8,000 - $9,000 and she hopes that her earnings will eventually surpass her expenses. Jennifer would like to treat her bicycle racing activity as a business (making the loss resulting from the activity in 2017 deductible on their tax return) rather than continuing to treat the activity as a hobby (limiting their ability to deduct expenses associated with this activity). ISSUE
1. Is bicycle racing a for-profit activity for Jim and Jennifer’s 2017 tax return?
63
[Reviewer] PREPARER CONCLUSION [Reviewer_business] The Roes' activity should be treated as a business. [Reviewer_omitted] Note: For this case study the preparer’s conclusion has been purposely omitted. ANALYSIS Treasury Regulation §1.183-2 provides guidance regarding when an activity is considered to be carried on for profit (deductions allowed under sections 162/212), including nine factors for consideration. Activities are not required to satisfy all nine factors; however, the courts apply these factors when determining whether an activity is engaged in for profit. Each of the nine factors is applied to Jennifer’s bicycle racing activity below. The activity is operated in a businesslike manner. Jennifer keeps track of her earnings and expenses with businesslike records. She has a thorough business plan and plans to continue improving her skills so she can improve her profitability. Taxpayers have the required expertise and/or have consulted appropriate advisors. Jennifer is ranked in two events, which suggests expertise. Taxpayers devote time and effort to the activity that is appropriate to a serious venture. There is no set amount of time required to satisfy this factor; however, Jennifer devotes a significant amount of time to the activity. She trains daily, which could be viewed as regular exercise but may be viewed as business activity. In addition, she adjusts her work schedule at the hospital during racing season to allow for further dedication to the sport. Taxpayers with reasonable expectations of asset appreciation can use that expected appreciation as future profit. This factor is not applicable to the current activity as Jennifer’s only asset is the bicycle and trailer, which will both depreciate in value. Taxpayers have demonstrated success in previous business ventures. Jim and Jennifer are both successful in their careers. A review of past returns shows that the Roes’ have started businesses before; however, none of those businesses remain in operation.
64
The taxpayer has a history of income or losses with respect to with the activity. Jennifer’s racing has produced losses over the past three years. The activity has generated occasional profits and may become profitable. Jennifer’s racing has generated sponsorships and some winnings. She expects her winnings to become more frequent, which could produce profits in the future. The taxpayer does not have sufficient income from other sources to offset losses. Jim and Jennifer are already high-income taxpayers; they are not dependent on her racing activity as a source of income. The taxpayer derives elements of recreation or physical pleasure. Although she maintains businesslike records and expects to eventually make a profit, Jennifer derives significant recreation and physical pleasure from her bicycling.